Google AI Mode: SEO guide to conversational search | Ighenatt

Conversational search does not start when a user types a long question. It starts when Google decides that one query is not enough to answer it. That is the operational shift behind AI Mode: you are no longer optimising a single SERP, but a chain of exploration.

For a dental clinic in Barcelona, the query “best treatment to straighten teeth without metal braces” is no longer just an informational keyword. In AI Mode it can open subqueries about Invisalign, clear aligners, average prices in Spain, patient age, review appointments, financing, and local reputation. Then come the follow-ups: “how long does it take?”, “does it hurt?”, “are there clinics near Gracia?”. If your content only answers the first phrase, it disappears in the second round.

Google introduced AI Mode as an experiment in March 2025, expanded the roadmap at I/O 2025, and has continued adding layers such as Search Live and AI Mode in Chrome. This article does not frame April 22, 2026 as an official launch. It is a working guide for SEO teams deciding what to change in content, internal linking, measurement, and prioritisation.

What changes when Google searches by conversation

AI Mode is useful when intent requires exploration, reasoning, or comparison. Google explains it with questions that previously required several searches: comparing options, understanding a concept, or planning a task with constraints. The user does not write a perfect keyword; they describe a problem with nuance.

That changes three parts of SEO work. The first is topical architecture. A page for “clear aligners Barcelona” is not enough if the surrounding site does not answer duration, price, types of aligners, adult cases, appointments, financing, and medical trust signals. The second is internal structure: Google Search Central recommends that important content be textual, crawlable, and easy to discover through internal links. The third is passage citability, which connects directly with the evolution of GEO and SEO.

The useful analogy is not “SEO for AI”. It is a well-prepared sales conversation. A good consultant does not memorise one answer; they anticipate objections, bring documentation, and know what to show when the client asks something else. AI Mode behaves in a similar way: it opens branches, keeps part of the context, and looks for pages that can support those branches.

In practice, audit each priority page with a four-column table: initial query, probable subqueries, follow-ups, and required assets. For an ecommerce store selling running shoes in Valencia, “running shoes for plantar fasciitis” can branch into cushioning, drop, insoles, pronation, size availability, returns policy, and physical stores. Each subquery needs a verifiable answer, not a generic paragraph.

AI Mode, AI Overviews, and Search Live are different products

Mixing up the names leads to bad decisions. AI Overviews are generated answers that appear on the results page when Google believes they help users understand a complex question. AI Mode is a separate conversational experience designed for deeper follow-ups and supporting web links. Search Live is a way to use AI Mode with voice and, according to Google announcements, progressively with visual capabilities.

Google Search Central adds an important nuance: AI Mode and AI Overviews may use different models and techniques, so the links they show may vary. That is uncomfortable for SEO, but necessary. Appearing in an AI Overview does not guarantee appearing in AI Mode. Being present in AI Mode does not mean the classic AI Overview block will cite the same URL.

The operational distinction looks like this:

The contrarian point: not every page needs to become a conversation. A transactional “book an SEO audit” page may still work best with a clear offer, proof, and a form. Conversational work pays off more for topics with chained doubts: private healthcare, education, tourism, B2B software, product comparisons, legal services, and medium-to-high-consideration purchases.

According to Soufi Esmaeilzadeh, Product Management Director at Google Search, AI Mode was designed so people can express complex questions in natural language instead of relying on minimal keywords. Translated into operations: write for how the real customer asks, not for how a keyword tool exports a term.

Query fan-out turns one keyword into a subquery map

Query fan-out is the most important mechanism for SEO teams. Google describes it as issuing multiple related searches in parallel across subtopics and data sources. At I/O 2025, Google explained that AI Mode breaks a question into subtopics and launches many simultaneous searches to go deeper than a traditional Google search. Deep Search pushes the same logic further, with hundreds of searches for cited reports.

You do not need to know the exact internal queries Google runs. You need content that survives that breakdown. Use this workflow:

1. Start with a high-value query. Example: “best invoicing software for freelancers in Spain”.
2. List predictable branches. Monthly price, Spanish bank integrations, e-invoicing, Verifactu, Spanish-language support, connected accounting firm, mobile app, migration from Excel.
3. Assign each branch to a URL or section. Not every branch deserves its own page; some work as H2s, FAQs, or comparison tables.
4. Add local proof. BOE, Agencia Tributaria, provider documentation, euro pricing, and Spanish freelancer conditions.
5. Connect branches with internal links. If the main page does not link to secondary answers, fan-out may find someone else first.

This work resembles the analysis of Perplexity and ChatGPT for SEO, with one key difference: AI Mode draws from Google’s index and quality systems. That means fundamentals still matter. The page must be indexed, snippet-eligible, and not hide its core answer behind JavaScript, opaque accordions, or PDFs without equivalent HTML.

A practical case: a vocational training school in Barcelona wants to capture “cybersecurity course with internships”. The ideal page does not only sell the course. It answers career paths in Catalonia, required technical level, whether internships are included, course duration, certifications, study formats, and financing. Each answer has 60 to 120 words, a source when needed, and a link to more detail. That structure does not guarantee visibility, but it gives AI Mode pieces it can recombine.

Follow-ups require content for the second and third question

Most SEO briefs stop at the first query. AI Mode becomes interesting right after it. Google has shown follow-ups in AI Mode and, in January 2026, announced a smoother transition from AI Overviews into AI Mode conversations. That forces content teams to design for user memory.

Think about a travel query: “weekend getaway on the Costa Brava without a car”. The first answer needs destinations and routes. The first follow-up might be “what if it rains?”. The second might be “any hotel with spa and dinner?”. The third: “can I do this from Barcelona Sants on Friday?”. If your hotel website only has a pretty landing page, it loses. If it has guides on public transport, rainy-day plans, maps, timetables, FAQs, and booking links, it competes better.

The concrete tactic: add “next question” modules to content that already gets traffic. You do not need to label them that way on the page, but you do need to cover those doubts. Detect them through Search Console, People Also Ask, sales calls, support chats, Google Business Profile reviews, and real conversations with the sales team. At Ighenatt we usually divide follow-ups into three groups: clarification (“what does it mean?”), comparison (“which option fits me?”), and action (“what do I do now?”).

For teams with limited time, start with 20 URLs. Export the pages with the most impressions, select those with informational or comparative intent, and add four to six short blocks that answer follow-ups. Each block should stand alone. One answer sentence, two context sentences, one proof point or source, and one internal link. No filler. We have seen 3,000-word articles fail because they do not answer a single concrete question.

Search Live and Chrome AI Mode change the consumption context

Search Live brings voice into AI Mode. Google first announced it for Labs users in the United States, with audio answers, back-and-forth conversation, on-screen links, transcripts, and continuity from history. Google also said Search Live uses a version of Gemini with voice capabilities and applies query fan-out to show a broader set of web content.

This matters for Spanish businesses because voice changes how people ask. Nobody says “best labour consultancy Madrid price”; they say “I need a labour consultancy in Madrid to help with contracts and payroll, what should I check?”. The useful answer needs to be clearer, less ornamental, and more decision-oriented. Overly creative headlines, tables without context, and vague FAQs lose value in a spoken interface.

Chrome AI Mode adds another twist. The April 16, 2026 update lets users explore pages side by side with AI Mode, ask follow-ups without losing context, and add recent tabs, images, or PDFs into the search. For SEO, this turns the webpage into conversational material. The user may be reading your comparison, open a competitor next to it, and ask “which fits a 12-person SME better?”.

The preparation is concrete. Use honest comparison tables. Explain limitations. Show visible dates. Name entities Google can recognise: neighbourhoods, cities, regulators, tools, brands, and regulations. If a tax advisor serves freelancers in Spain, mention “modelo 303”, “factura electronica”, “Verifactu”, and “AEAT” when relevant. If a SaaS sells to hotels, explain PMS integrations, channel managers, and booking engines. AI Mode does not reward vague language.

Measurement and priorities for the Spanish market

Google Search Central says links shown in AI Mode and AI Overviews are included in Search Console search traffic under the Web type. It also recommends combining Search Console and Google Analytics to understand how users arrive and what they do afterwards. The Performance report still shows clicks, impressions, CTR, and average position; the problem is that those metrics do not tell the full conversational story.

Measure by clusters and quality. If a guide on “e-invoicing for freelancers” gains impressions but loses CTR, do not automatically conclude that it got worse. It may be appearing in AI experiences that solve part of the doubt before the click. Check whether assisted conversions, time on page, form leads, branded queries, and sessions entering through secondary content increase.

Connect this view with your GEO metrics for AI visibility. Build a monthly panel with five signals: queries that trigger AI answers, URLs cited or linked, follow-ups where you appear, competitors cited, and downstream conversion. You do not need a perfect tool to start. A manual sample of 30 critical queries, run every month with screenshots and notes, already reveals patterns.

The priority in Spain should be pragmatic. First, indexable and linked content. Second, clusters with answers to real subqueries. Third, verifiable local data. Fourth, conversion tracking. Fifth, format experiments: tables, FAQs, comparisons, decision modules, transcribed videos, and images with textual context.

The operational step for today: choose one URL that already generates business, not one that only gets traffic. Write the initial query and ten follow-ups. Mark which ones the page answers clearly. Where there are gaps, add short sections, sources, and internal links. Then repeat the exercise monthly. AI Mode does not ask for magic; it asks your content to support a real conversation.

Frequently asked questions about AI Mode and conversational search

Is AI Mode officially available in Spain?

AI Mode features have rolled out in phases, and many of the updates Google described started in the United States or Labs. For the Spanish market, the practical move is not to assume immediate universal availability, but to prepare content and measurement for the conversational pattern Google is gradually bringing into Search, AI Overviews, Search Live, and Chrome.

Does AI Mode replace AI Overviews?

No. AI Overviews are snapshots that appear when Google decides they add value inside the classic SERP. AI Mode is a conversational search experience for longer, comparative, or exploratory questions. They can connect: Google has already enabled some users to move from an AI Overview into an AI Mode conversation.

Do I need special schema, llms.txt, or an AI file to appear in AI Mode?

Google Search Central says there are no additional technical requirements and no special schema markup for appearing in AI Mode or AI Overviews. The essentials are that the page is indexed, eligible to show with a snippet, crawlable, textual, useful, reliable, and easy to find through internal links.

How do I measure whether AI Mode is affecting performance?

Google includes AI Mode and AI Overviews links in Search Console Web traffic, but it does not always provide a clean segmentation by experience. Combine Search Console, GA4, manual tracking of conversational queries, keyword clusters, and business metrics such as leads, assisted sales, and session quality.